Dense 3D reconstruction using a rotational stereo camera

Abstract

This paper proposes a method to construct a three-dimensional (3D) model of an outdoor scene containing a large amount of information by combining local stereo images which are acquired from multiway. It is difficult to reconstruct outdoor scenes accurately using a single view of a stereo camera because the distance to a target is very far compared with the stereo camera's baseline length. Therefore the proposed method models the uncertainty of a 3D point measured by the stereo camera. And this method makes the 3D model accurately by reducing the uncertainties of the 3D point from various angles to the region which includes the correct point. Multi-view stereo images are captured using a rotational stereo camera that swings back and forth and can capture not only the upper surface of an object but also the side surface of the object. As an experiment, a blimp robot with a rotational stereo camera is used to capture aerial stereo images of the ground. The proposed method achieves the accurate reconstruction of the outdoor 3D model using our method.